INDEX

Explanations

aspects of life and society

The neuron is triggering on prominent, information-bearing content words (e.g. key nouns like “work,” “food,” “banking,” “name”) – in other words, it flags the document’s most distinctive topical terms.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

it

-1.84

сыз

-1.70

 costruire

-1.59

ШЕ

-1.55

 inserire

-1.48

 несмотря

-1.45

 déclare

-1.43

 virtù

-1.43

を行う

-1.41

 risultati

-1.40

POSITIVE LOGITS

and

5.97

piram

1.64

dekor

1.63

or

1.60

 zonder

1.59

1.51

',)

1.49

kontrol

1.43

komfort

1.41

แรก

1.41

Activations Density 0.872%